COINVENT - Concept Invention Theory
Funded by the European Commission FP7 programme
- Artificial Intelligence Research Institute (IIIA-CSIC), Barcelona (Coordinator)
- University of Edinburgh, U.K.
- University of Osnabruck, Germany
- Universty of Bremen, Germany
- Goldsmith's College, London, U.K.
- University of Dundee, U.K.
- Aristotle University of Thessaloniki
Members of AUTH research team
- Emilios Cambouropoulos (School of Music Studies - principal investigator)
- Costas Tsougras (School of Music Studies, AUTH)
- Danae Stefanou (School of Music Studies, AUTH)
- Leontios Hadjileontiadis (Dept of Electrical and Computer Engineering, AUTH)
- Christina Anagnostopoulou (School of Music Studies- University of Athens)
- Maximos Kaliakatsos-Papakostas (School of Music Studies, AUTH - postodoctoral research fellow)
- Asteris Zacharakis (School of Music Studies, AUTH - postodoctoral research fellow)
The overall aim of COINVENT is to develop a computationally feasible, cognitively-inspired formal model of concept creation, drawing on Fauconnier and Turner's theory of conceptual blending, and grounding it on a sound mathematical theory of concepts. To validate the model, a proof of concept of an autonomous computational creative system will be implemented that will be evaluated by humans in two testbed scenarios: mathematical reasoning and melodic harmonization.
The specific music-related component of COINVENT is to study how a melodic harmonisation assistant may be creative, not only in a sense of generating new harmonisations in given styles, but, rather, generating new harmonic concepts that result in new harmonic spaces per se. Having devised basic melodic harmonisation systems (e.g. employing statistical grammars), say, for tonal/modal/atonal music, the question is how a melodic harmoniser may be prompt to invent new harmonic concepts when dealing with potential conflicts or purely for creative purposes.
The aim of this specific project is:
- To study harmonic blending and concept invention in the context of a melodic harmonisation assistant.
- To analyse and describe formally ways learned harmonic probabilistic models may be blended/transformed at different hierarchic levels to create new harmonic forms and styles.
- To develop and evaluate a creative melodic harmonisation assistant that may invent new harmonic concepts and create new "blended" harmonic spaces that are appropriate both for tonal and for nonstandard non-tonal melodic styles and may produce original novel harmonic progressions.
This study led to the development of the CHAMELEON melodic harmonisation system. For more information please visit the CHAMELEON webpage: